Red-Handed: Sherlock Holmes Deduction Puzzles

Something happened in a small bay town, and every suspect has an account of where they were. All but one of those accounts is true. The one who lied is the one who did it, and they were alone at the scene when it happened — which is exactly why they needed a story. It is the oldest shape in detective fiction, and it is solved the oldest way: when you have eliminated the impossible, whatever remains is your answer.

You solve it by rebuilding the evening. Every suspect goes into a chair, and every object goes into somebody's hand. Two things are being worked out at once, and they lean on each other: knowing who carried the crowbar often tells you which room they were in, and knowing who shared a room often tells you who could not have been carrying it. That interlock is what makes the deductions chain rather than pile up.

The statements never simply hand you an answer. Instead of naming somebody, a forensic line describes them — whoever carried the knife is left-handed — and you resolve the description against the printed cast sheet. Every suspect carries their traits openly, so the inference is always answerable from what is on the screen, never from memory or luck.

The one liar is a solving tool, not just a twist. Once two accounts contradict each other, you know the liar is one of those two — which means everybody else is telling the truth, and you can use their accounts as hard facts. Most solvers find that step more useful than any single clue in the case. It is also the move that makes these feel like Holmes rather than a quiz: you are not hunting for a hidden fact, you are testing suspects one at a time and discarding the ones the evidence will not allow.

Red-Handed sits alongside Deduction Puzzles, Logic Grid Puzzles, and Brain Teasers in our logic collection. Where a classic logic grid asks you to match categories, Red-Handed asks you to decide which statement to throw away — and that single extra question is what turns a matching exercise into a case.

Read the forensics first. Those lines come from the investigators, not the suspects, so they are always true. They are the only statements you can build on without checking.

Look for two accounts that cannot both hold. Three people claiming the same two-chair room, or somebody vouching for a person who says they were alone. One of them is lying, and the rest of the town is not.

Test your candidates. Suppose one of them lied. They were alone at the scene. If another suspect's account — true, because only one person lies — puts a second body in that room, that suspect is cleared and you have your answer.

Nothing needs a guess. Holmes called guessing a shocking habit, destructive to the logical faculty, and no case here will ever ask you for one. Every puzzle has exactly one arrangement that fits, and a separate program proves it is reachable by reasoning alone before the case ships. If you are stuck, there is a deduction you have not made yet — not a coin to flip.

Difficulty here is the length of the chain, not the size of the board — and that is a measured claim, not a slogan. Widening the cast actually makes cases easier: more suspects means more ways for the evidence to sit consistently, so each statement rules out less and the reasoning bottoms out sooner. A case where nothing can be worked out until three other things have been worked out is the hard one. Every puzzle is graded by how many rounds of deduction it takes, and the tiers grow that number first.

Easy gives you four suspects and a short chain — enough to learn the shape of the reasoning. Medium and Hard add a suspect and take facts away, so more of the board has to be inferred rather than read. Expert keeps Hard's six suspects and four rooms on purpose and lengthens the chain instead: the same board, a deduction further from the surface. Sherlock, the top tier, is the longest reasoning in the collection — named for the detective because that is the standard it is held to, first line to last.